Visa, Mastercard and Ant International Plan a Shared Trust Framework for AI Shopping Agents

AI agents are moving from recommending products to potentially completing purchases for people. That raises a basic payments question: how can a merchant, wallet, acquirer or card network tell whether an agent is legitimate, who operates it and whether it is permitted to act?

On September 10, 2026, Visa, Mastercard and Ant International said they had begun collaborating on a Know-Your-Agent (KYA) interoperability framework. The proposed framework is intended to streamline the identification and onboarding of AI agents across card networks, digital wallets, agent platforms and online marketplaces. It would allow participants to recognize trusted agents across ecosystems while preserving each participant’s own approval, verification and risk-management processes. 16

What the KYA initiative is—and is not

KYA is a collaboration to develop common principles for agent identity and onboarding. It is not an announced, ready-to-deploy universal product or a single central approval system. The partners have said that each network will retain its own verification and decisioning process. 112

That distinction matters. Interoperability could reduce fragmentation for an agent that needs to operate across wallets, marketplaces and payment rails, but it does not mean an agent approved in one context automatically receives approval everywhere else.

How KYA builds on the companies’ existing protocols

The three companies already have separate approaches to agentic commerce:

• Visa: Trusted Agent Protocol
• Mastercard: Verifiable Intent
• Ant International: Agentic Mobile Protocol

The KYA work is intended to align these approaches around common cross-network principles rather than replace them. Ant International has also been rolling out its Agentic Mobile Protocol through wallet and acquirer ecosystems. 612

In practical terms, the effort is about making an agent’s identity and operating context easier to assess when it crosses payment ecosystems—while leaving final trust and risk decisions to the relevant network or participant.

The three proposed pillars

Reporting on the proposal identifies three core areas of work: 3

1. Cross-network operator traceability
   Participants would be able to identify the organization operating an AI agent and establish accountability as that agent moves between systems.

2. Shared certification requirements
   The partners would explore common expectations for agent eligibility and onboarding, helping ecosystems assess whether an agent meets baseline requirements.

3. Continuous transaction monitoring
   Trust would not depend solely on a one-time onboarding event. Monitoring agent behavior and transactions over time is intended to support ongoing risk management.

Together, these pillars address a central problem in agentic payments: a payment system needs visibility not only into the payment credential, but also into the software acting for a customer and the authority under which it acts.

Why payment companies see a large opportunity

The companies have pointed to projections that AI agents could orchestrate $3 trillion to $5 trillion in global consumer commerce by 2030. That is a projection, not a forecast of KYA adoption, transaction volume or revenue. 1114

Mastercard separately predicts that more than one in 10 online shoppers could routinely use AI agents to shop and pay by 2030. Its report suggests repeatable, lower-effort purchases—such as groceries, medicines and subscriptions—may be among the earliest use cases. 2021

Mastercard-commissioned research covering 26,000 parents and teenagers aged 13 to 18 across 13 countries found that teenagers use AI to support shopping decisions at roughly twice the rate of parents. 2427

One figure needs care: the widely repeated 62% statistic refers to young people in the UK who said AI would significantly change how their generation shops. It should not be presented as a result for all teenagers in the 13-country study. 24

What still has to be solved

The KYA announcement is an early infrastructure step, not evidence that agentic commerce has reached broad deployment. A common recognition framework may help, but merchants, issuers, consumers and regulators still need confidence in several areas:

• whether the agent is correctly identified;
• whether the consumer gave explicit consent;
• what the agent is authorized to buy and under what limits;
• who is liable when an agent makes an unwanted, fraudulent or disputed transaction; and
• how suspicious behavior is detected and stopped.

Mastercard’s own research emphasizes the need for clear frameworks covering identity, consent, authorization and liability. 21 No implementation timetable or merchant-pilot program was included in the September 10 announcement, and the framework itself remains to be developed. 16

The bottom line

Visa, Mastercard and Ant International are trying to establish a common trust layer for AI agents that shop and pay across different payment ecosystems. The proposed KYA framework could make agent identities more portable and traceable, but it deliberately does not override the separate approval and risk controls of participating networks. Its success will depend on whether the industry can turn shared principles into deployable rules for identity, consent, monitoring and liability. 136